DOVE diagnostic implementation plan
A. I Configuration storage medium
A.3 Configuration information description and management
The configuration utility
There will be a floppy- or Ethernet-bootable configuration utility that will allow initializing and modifying the configuration EEPROM. Setting the EEPROM is a manual, menu-driven process. The configuration utility will not be able to do it automatically, because the hardware is not completely self-identifying. This utility will be based on UtilityPilot and written in Mesa. It will be similar to the disk initialization utility, and could possibly be incorporated into that utility.
A safe default configuration
The management of the configuration EEPROM must be carefully thought out to avoid any circular dependencies. One such circular dependency could occur if the EEPROM were
required to be initialized before the configuration utility program could be booted to initialize it. The strategy to avoid this p'redicament is to substitute a safe default configuration whenever the EEPROM is found to be invalid or uninitialized. The default configuration, must allow the configuration utility to be booted and run.
The important parameters of the safe default configuration are memory size, display, floppy disk, and Ethernet. It should be safe to set the memory to the smallest allowable size, the display to nexperiment/' the floppy disk to Uboth types present," and the Ethernet to present. Note that two backplane signals will identify the processor type (currently either DAYBREAK or DAISY).
For the display, ttexperiment" instructs the
rop
boot code to experiment with the display controller to determine which crystal oscillator is installed and thus which type of display is expected. This involves timing the interval between vertical retrace interrupts. For each supported display, this interval will be different. Once the display type has been identified, the display controller can be correctly initialized to allow use of the display by the configuration utility.For the floppy disk, "both types present" means that both types of floppy disks should be assumed to be present. This would allow the user to boot from whichever type of floppy is present, if there is one.
An alternate strategy for the floppy would be to reserve a fixed location in a sector on cylinder 0, head 0, of every floppy diskette for floppy drive type information. This would require that all DOVE floppies be formatted identically on the track on cylinder
°
underhead 0, so that no knowledge of the floppy drive t-ype would be required to read that track.
The safe default configuration should allow a configuration utility program to be booted from either type of floppy or Ethernet and use the display, regardless of type. This strategy requires the ability to distinguish invalid and uninitialized EEPROMs from valid EEPROMs. By including a word of checksum followed by its complement in the EEPROM, it should be virtually impossible for an uninitialized or otherwise invalid EEPROM to appear valid.
Managing the EEPROM contents
Managing the contents of the configuration EEPROM can be divided into three phases:
1. setting it initially when the machine is configured prior to shipping.
2. reading it during system initialization while booting.
3. changing it whenever the hardware configuration (or other EEPROM information) is changed.
Setting it before shipping
When a DOVE workstation is configured prior to shipping, the configuration utility should be run to set the EEPROM correctly. This is when the safe default configuration will be used most.
Reading it during booting
DOVE Program Diagnostic Strategy Version 5.0
A
will verify the checksum and its complement in the EEPROM before trusting it. If the checksum or complement is incorrect, the booting code will post an error indication in the LED status indicators. A keystroke from the user is required to continue after an EEPROM error is indicated. This will cause the booting code to try to complete the booting process using the safe default configuration. The user should try to boot the configuration utility and attempt to reset the configuration EEPROM, assuming the checksum problem is only transitory. If the problem is persistent, then the lOP (the FRU containing the EEPROM) will have to be replaced, and the new EEPROM reset.
There should be a way to force the booting code to use the safe default configuration. This would be necessary, for instance, if the EEPROM checksum and complement are valid but the indicated configuration is set incorrectly. It should be possible to circumvent the EEPROM manually by holding down a special key during booting.
Changing it when the hardware is changed
When the hardware configuration is changed, the configuration utility should be run to change the EEPROM appropriately. It should be possible to change the EEPROM before"
changing the hardware, if that is ever required, since the EEPROM will probably be read during initialization only.
Rigid disk backup
If possible, the EEPROM configuration information should be backed up on the rigid disk.
This facilitates restoring the configuration information to the EEPROM after replacing the IOP/RDC board, which contains the EEPROM.
-It should be possible to reserve a sector on the rigid disk for backing up the EEPROM contents. A good candidate is the sector where we currently put the first page of the initial microcode. The initial microcode could start on the following page on DOVE machines, since this location is allowed to be processor-dependent. It should be possible to read this page without knowing the type of disk. If not, a less attractive alternative would be to require the user to manually specify the type of the disk to the configuration utility.
Note that it would probably be unwise to store the configuration information only on the disk, and not include an EEPROM on DOVE. One reason is the significantly lower reliability of disk memory compared to EEPROM memory. Couple that with the fact that there would then be no place to back up the configuration information, and reliable maintenance of this information could become a significant problem. Replacing the disk would result in the loss of the information. A third reason is that it may not be possible to read a particular sector on all disks without knowing the disk type. An EEPROM or other storage medium other than the disk is required if the latter is true, because it would be unacceptable to ask the user the disk type on every hard boot.
Contents of configuration EEPROM
The configuration EEPROM should include the following kinds of information:
• display type
• keyboard type
• rigid disk presence and type
• floppy disk presence and type
• main memory size
• virtual memory size
• Ethernet presence
• RS-232-C channel A and B device types and attributes
• option board types and attributes, including PC emulation
• booting options (default boot device, whether to run Boot Diagnostics)
• memory lock mode (hardware or software)
• address of last main memory parity error and count
• list of bad main memory pages
• checksum and its complement